Dear Organizers,
Based on my log output, it was found that there were 2112 samples and 248 unique patients in the test set.
However, only 248 predictions are needed (for each patient).
So whether the patient IDs are in the same order may affect the final result, which is fine if matching the same patient IDs has been taken into account in the backend validation algorithm for the calculation. I would just like to know if the possibility of such a problem exists.
Created by Zhixiang Lu Leo_Lu Problem solved, I just checked https://github.com/Sage-Bionetworks-Challenges/TB-DREAM-Challenge-Infra/blob/main/docker/score.py and the id is matched in the algorithm.
```
def main():
"""Main function."""
args = get_args()
pred = pd.read_csv(args.predictions_file)
gold = pd.read_csv(args.goldstandard_file)
df_combine = pred.merge(gold, on=['participant'], how='left')
scores = score(df_combine)
with open(args.output, "w") as out:
res = {
"submission_status": "SCORED",
**scores
}
out.write(json.dumps(res))
```
Drop files to upload
Questions about the number of test sets and the order of patients IDs page is loading…